Page 1 of 3
TCEC Season 6 - Stage 1c now live
Posted: Tue Mar 04, 2014 11:38 am
by Martin Thoresen
Congratulations to Naum, Stockfish, Fire, Shredder and Spike for qualifying for Stage 2 after Stage 1b ended late last night.
Equinox is still the best 6th placed engine so unless the 6th best engine in 1c can score at least 7 points, it will qualify for Stage 2 as well.
Code: Select all
N Engine Elo Pts SB Na St Fi Sh Sp To Ni Tg Ex Ar Ne Ja
1 Naum 4.6 3053 9.0 43.25 = = 1 = 1 1 1 = 1 1 1
2 Stockfish 180214 3181 9.0 43.00 = 1 = = = 1 1 1 1 1 1
3 Fire 3.1 3102 8.0 35.25 = 0 = 1 = = 1 1 1 1 1
4 Shredder 12 2966 6.5 31.25 0 = = = 1 = 1 0 1 1 =
5 Spike 1.4 2935 6.5 30.75 = = 0 = = 1 = = 1 = 1
6 Tornado 5 2809 6.0 26.50 0 = = 0 = = 1 = 1 = 1
7 Nightmare 7.8 2880 6.0 23.75 0 0 = = 0 = 1 1 = 1 1
8 Toga II 010214 2759 4.5 13.75 0 0 0 0 = 0 0 1 1 1 1
9 Exchess 7.18b 2696 4.0 19.50 = 0 0 1 = = 0 0 0 = 1
10 Arasan 16.3 2466 3.0 10.00 0 0 0 0 0 0 = 0 1 1 =
11 Nebula 2 2097 2.5 9.25 0 0 0 0 = = 0 0 = 0 1
12 Jazz 799 2371 1.0 4.75 0 0 0 = 0 0 0 0 0 = 0
All the games can of course be found in the Archive page and downloaded from there:
http://tcec.chessdom.com/archive.php
Stage 1c:
16 cores, ponder off. Up to 16 GB hash. Time control is 120' + 30".
Single round robin.
Top 5 qualifies for Stage 2 (+6th best placed in 1a, 1b and 1c combined)
http://tcec.chessdom.com/live.php
Full rules can be read on the TCEC website, click "Help" in the top menu then "Rules & Information".
Code: Select all
N Engine Elo Pts SB Pr Sc Ry Sp Th Oc Al Ko Ch Jo Ju Cr
1 Protector 1.6b12 2928 0.0 0.00
2 Scorpio 2.77 2769 0.0 0.00
3 Rybka 4.1 3131 0.0 0.00
4 Spark 1 2800 0.0 0.00
5 The Baron 3.36 2661 0.0 0.00
6 Octochess 5374 2367 0.0 0.00
7 Alfil 14.1 2335 0.0 0.00
8 Komodo 1188 3177 0.0 0.00
9 Chiron 2 3008 0.0 0.00
10 Jonny 6 2930 0.0 0.00
11 Junior 13.3 2977 0.0 0.00
12 Crafty 23.8 2482 0.0 0.00
Re: TCEC Season 6 - Stage 1c now live
Posted: Tue Mar 04, 2014 10:50 pm
by Isaac
In the game Scorpio-Crafty, one can see that for most moves, both engines reached exactly 2147.5 M nodes and then made their move instantly.
Is that a coincidence, a bug or something else? I think that the probability for this to happen is lower than 1 divided by 1 mole.
Re: TCEC Season 6 - Stage 1c now live
Posted: Tue Mar 04, 2014 11:11 pm
by Martin Thoresen
Isaac wrote:
In the game Scorpio-Crafty, one can see that for most moves, both engines reached exactly 2147.5 M nodes and then made their move instantly.
Is that a coincidence, a bug or something else? I think that the probability for this to happen is lower than 1 divided by 1 mole.
Interesting observation. This behavior seems to be prominent in the start and middle of the game only. I have no good explanation other than both of these engines are using the Wb2Uci adapter which to me implies this as a source for these numbers.
Re: TCEC Season 6 - Stage 1c now live
Posted: Wed Mar 05, 2014 12:03 am
by michiguel
Martin Thoresen wrote:Isaac wrote:
In the game Scorpio-Crafty, one can see that for most moves, both engines reached exactly 2147.5 M nodes and then made their move instantly.
Is that a coincidence, a bug or something else? I think that the probability for this to happen is lower than 1 divided by 1 mole.
Interesting observation. This behavior seems to be prominent in the start and middle of the game only. I have no good explanation other than both of these engines are using the Wb2Uci adapter which to me implies this as a source for these numbers.
It must be maxing out at 2^31, which is 2147483648 nodes, the size of an integer in 32 bits. I was thinking of writing an xbii2uci adapter to experiment (XB protocol II to UCI) and may end up doing it...
Miguel
Re: TCEC Season 6 - Stage 1c now live
Posted: Wed Mar 05, 2014 12:15 am
by Isaac
Thank you guys. That is funny
Does this mean that they cannot think any further than that number of nodes? I have checked all moves and there is no move with a greater number of nodes than 2147.5 millions.
Re: TCEC Season 6 - Stage 1c now live
Posted: Wed Mar 05, 2014 12:23 am
by Martin Thoresen
Isaac wrote:Thank you guys. That is funny
Does this mean that they cannot think any further than that number of nodes? I have checked all moves and there is no move with a greater number of nodes than 2147.5 millions.
No, the adapter comes in
after the engine output. So it has no impact on performance nor does it set any limits of any kind. It simply means that if the actual node number is higher than 2147.5 then the adapter simply rounds down to this number before sending this value to cutechess.
Re: TCEC Season 6 - Stage 1c now live
Posted: Wed Mar 05, 2014 12:25 am
by michiguel
Isaac wrote:Thank you guys. That is funny
Does this mean that they cannot think any further than that number of nodes? I have checked all moves and there is no move with a greater number of nodes than 2147.5 millions.
I do not think an adapter will influence when an engine should stop thinking based on nodes, unless the implementation is really screwy (is that a word?), but it could just report that maximum no matter how many nodes it receives. I am just speculating.
Miguel
Re: TCEC Season 6 - Stage 1c now live
Posted: Wed Mar 05, 2014 3:57 am
by Isaac
Thanks once more guys!
Re: TCEC Season 6 - Stage 1c now live
Posted: Wed Mar 05, 2014 12:03 pm
by Martin Thoresen
michiguel wrote:
It must be maxing out at 2^31, which is 2147483648 nodes, the size of an integer in 32 bits. I was thinking of writing an xbii2uci adapter to experiment (XB protocol II to UCI) and may end up doing it...
Miguel
That would be great Miguel. I would gladly help you test it. I was not aware of the limitation of Wb2Uci before, actually. :S
Re: TCEC Season 6 - Stage 1c now live
Posted: Sun Mar 09, 2014 1:18 pm
by Martin Thoresen
Standings after 5 of 11 rounds:
Code: Select all
N Engine Elo Pts Gm SB Pr Ju Ko Ry Ch Sc Sp Th Jo Al Oc Cr
1 Protector 1.6b12 2928 4.5 5 10.00 = 1 1 1 1
2 Junior 13.3 2977 4.0 5 8.00 = 1 = 1 1
3 Komodo 1188 3177 4.0 5 8.00 = 1 = 1 1
4 Rybka 4.1 3131 3.0 5 6.25 = 0 1 = 1
5 Chiron 2 3008 2.5 5 5.50 0 0 1 = 1
6 Scorpio 2.77 2769 2.5 5 5.00 0 = 1 0 1
7 Spark 1 2800 2.5 5 2.75 0 0 1 = 1
8 The Baron 3.36 2661 2.0 5 4.25 = 0 = = =
9 Jonny 6 2930 1.5 5 4.00 0 0 = 1 0
10 Alfil 14.1 2335 1.5 5 3.00 0 0 = = =
11 Octochess 5374 2367 1.5 5 1.25 0 0 0 = 1
12 Crafty 23.8 2482 0.5 5 1.00 0 0 0 = 0